7 minute read 29 Jan 2019
Engineer programming robot in robotics research facility

How human-centered AI can help transform the energy industry

Artificial intelligence (AI) is humanly conceived but powered by technology. It stretches the possible and the limits of humans.

One of a suite of disruptive technologies that promises to transform our world is AI. It might sound very sci-fi (and yes, there is a lot of fancy stuff going on behind the scenes), but its real attributes are fairly rudimentary. It identifies patterns and applies logic in order to initiate an action – albeit faster and more precisely than humanly possible.

Take, for instance, an autonomous vehicle. The AI hidden behind the dashboard detects the lines in the road (the patterns) and applies reasoning (the logic) to determine whether to execute a maneuver (initiates the action).

Apply that same theory to the power and utilities sector, and AI can sift rapidly through mountains of data to identify patterns; it applies logic to determine how best to treat anomalies and initiates the appropriate action or remedy.

Far from replacing human ingenuity, AI complements it, while removing the tedium that would otherwise accompany data mining and the pursuit of valuable information.

AI-enabled efficiencies

Many of the traditional ways of working in the energy industry are ripe for an advanced technological intervention.

We might, for instance, use AI to empower chatbots in call centers so that the first few steps of customer contact are fully automated. Done well, this can deliver time and cost savings without adverse impact on the end-user experience. 

We could employ AI’s “deep learning” capabilities – an artificial neural network that analyzes different layers of information – to make better predictions about the maintenance of network assets so that intervention is timely but targeted. Just imagine the efficiency and cost savings that are possible when an AI solution identifies, with 99% certainty, that an overhead line really does warrant calling out a team of engineers to fix it.

Or, we could use AI to identify patterns of behavior that indicate customer dissatisfaction – perhaps tone of voice or choice of words or questions about energy usage or tariffs – enabling intervention and remediation to reduce frustration.

The fact is that customer expectations are being raised by disruptive innovations and other new world capabilities. They will not settle for substandard performance in billing or network maintenance or service. AI presents an opportunity to conquer hurdles that can impede customer satisfaction and quality service.

AI-enabled energy transition

We are fast approaching the point when energy is neither created nor consumed centrally. Energy “prosumers” will connect their distributed resources to the grid, downloading and uploading energy according to need, with potential payment for their surplus supply. 

Consumers will also connect their devices – including smart appliances – to the internet. More extensive technology take-up will mean even greater data generation. AI will sift through this data at a rapid rate, pinpointing patterns of behavior and making accurate predictions on energy demand. More data means deeper insights, which enable, in turn, an intelligent, stable and autonomous grid.

AI algorithms will, for instance, recognize patterns of behavior on, say, a weekday evening in 2025, when millions of EV drivers arrive home and put their vehicles on charge. By distinguishing between drivers who habitually use their cars overnight and those who leave vehicles charging until the following morning, the intelligent grid will ensure that the battery is sufficiently charged in time for the driver’s next journey, without exerting simultaneous load on the grid. 

AI is a big data game. By better understanding ownership of the data and how it can be shared and combined, meaningful algorithms can be developed to underpin trusted AI programs.

Accessing new revenue streams

For power and utilities companies seeking to access new business models and revenue streams, AI can help them to remain relevant and current beyond the energy transition.

They could, for instance, use AI to compress, analyze and monetize the huge swathes of data moving through the energy ecosystem. Or, they might follow the lead of technology startups by harnessing apps and other innovations to enhance the networked and connected home.

AI is a big data game. At EY, we are working with organizations to define their data architecture, data management and data governance. By better understanding ownership of the data and how it can be shared and combined, meaningful algorithms can be developed to underpin trusted AI programs.

AI challenges are human after all

Though AI enhances human capabilities artificially, many of its limitations are the consequence of human frailties.

Deep-learning AI algorithms, for instance, train themselves by sifting through large volumes of data. From this they learn to identify exceptions to the norm and to make reliable predictions. If utilities fail to cleanse and structure their data before letting AI loose on it, analysis and outcomes will be compromised.

Then there are issues relating to computer power – or rather, the lack of it. Some utilities are reluctant to migrate to cloud computing solutions due to concerns about data privacy and costs. It is, however, a prerequisite for AI given its extensive storage and processing needs. Failure to invest in computing power will impact the data-related insights that utilities can get from AI.

Utilities also have to come to grips with data privacy. They need to understand who owns the data, which data is confidential and how open data should be used and stored if they are to optimize its potential and comply with relevant regulations.

Some utilities recognize, however, that it takes input from both the IT function and the business itself to train a deep-learning network. Increasingly, we see engineers, shop floor workers, asset managers and program managers collaborate on AI capabilities. They jointly define and test a use case; they populate the system with relevant data, rather than drain the entire data pool; together, they deliver the right algorithm training so that data can be mined and its insights extracted.

Collaboration or partnership is a must for any utility. Otherwise, they could struggle with the level of technology sophistication and specialization that more nimble startups readily achieve.

Taking on or teaming up with competitors

AI is not strictly a solo endeavor. Collaboration or partnership is a must for any utility. Otherwise, they could struggle with the level of technology sophistication and specialization that more nimble startups readily achieve.

Some utilities incubate AI solutions in isolation, others collaborate, particularly with startups, to piggyback on their technology know-how.

Many look further afield, collaborating with startups in Germany, the UK, the US and the Middle East to access specialist capabilities. They work with omnichannel, intelligent customer support applications. These are essentially AI-powered chat solutions that understand customer conversations and automate repetitive processes, reducing resource needs and costs.

Some startups even offer platform architectures for storing, consuming and selling energy, while others work with utilities to deliver predictive maintenance solutions. By reducing unnecessary system intervention and delivering timely remedial action when it is needed, they focus costs and resources in all the right places.

How far can AI go?

Frankly, some utilities are slow off the mark; others show varying degrees of AI maturity.

While there is no need for utilities to invest huge sums right now, they must remain alert to startups, which have begun to roll out AI-enabled solutions that are smart – and which customers like. These fast movers will erode utilities’ conventional business models.

Looking ahead, a union between the Internet of Things (which offers a virtual environment through which distributed energy resources can be connected) and blockchain (which facilitates trusted transactions between buyers and sellers of homegrown electrons, without the intervention of a central authority) will reinvent energy delivery and trading.

Quantum computing – although some way off but attracting lots of investment – could be the big AI game changer. It will make deep-learning networks faster, more powerful and able to solve the trickiest challenges, all while storing even larger bodies of data.

Despite all the technology advances going on, many utilities are not riding the wave of innovation and risk losing some or all of their business to competitors. Pushing ahead with AI starts with:

  • Defining the AI strategy
  • Engaging the business on how to achieve AI transformation
  • Early experimentation, with startups or in-house innovation labs or accelerator programs
  • Pilots and test cases to understand what AI is and what it could do for the business

AI complements rather than replaces human intelligence. It doesn’t have all the answers to the future of energy; rather, it is the means by which answers might surface. It is the technology that enables tasks, particularly repetitive or labor-intensive ones, to be performed smartly and quickly, reducing costs and improving efficiencies. Those businesses that start adapting and testing AI technology right now stand a better chance of keeping up or working with the innovative startups that are shaking up the power and utilities industry.


By harnessing AI, the power and utilities industry will be able to turn volumes of customer data into meaningful insights that will contribute to a more secure and stable energy future.